A Retrospective Cohort Study of the GLA c.937G > T, p.Asp313Tyr Variant With No Evidence of an Association With Fabry Disease

Abstract

Background and Objectives

Fabry disease (FD) is a multisystemic disorder caused by pathogenic variants in the α-galactosidase (GLA) gene. There is ongoing debate on the disease-causing properties of the GLA variant NM_000169.3: c.937G > T, p.Asp313Tyr (D313Y). Like many others, we recently reclassified this variant from “variant of uncertain significance” to “benign.”

Methods

By leveraging Centogene's Biodatabank, we reassessed our classification and analyzed all samples (n = 38,242) molecularly tested for FD from Germany between 2015 and 2020 for GLA genotypes, clinical presentations, and biochemical phenotypes.

Results

The allele frequency of GLA variant p.Asp313Tyr in our cohort was increased compared with gnomAD v4.1.0 data for European (non-Finnish) population resulting in an odds ratio of 2.13 (95% CI 1.94–2.33; p < 0.0001). Compared with individuals without any GLA variant, there was no specific clinical nor biochemical phenotype in this group, supporting a diagnosis of FD.

Discussion

GLA variant p.Asp313Tyr is to be classified as a benign variant and does not cause FD. Because an enrichment of patients with this variant in specific high-risk populations has been seen in this cohort and by other groups, properly designed case-control studies are needed to clarify a potential role of p.Asp313Tyr as risk factor or modifier of a FD-independent dysfunction.

Introduction

Fabry disease (FD; OMIM #301500) is a rare X-linked metabolic disorder caused by pathogenic variants in the GLA gene (OMIM #300644), located on Xq22.1. This leads to α-galactosidase A (AGLA) deficiency, causing glycosphingolipid accumulation, particularly globotriaosylceramide and globotriaosylsphingosine (Lyso-Gb3), with consecutive multiorgan damage. In classic FD, male patients with minimal enzyme activity experience early symptoms such as neuropathic pain and hypohidrosis in childhood, progressing to renal, cardiac, and cerebrovascular complications. Variant FD forms involve higher residual AGLA activity with later onset and milder symptoms.¹ Unusually for X-linked recessive disorders, up to 70% of female heterozygotes are symptomatic.²

Genetic testing has revealed over 1,100 GLA variants,³ with no clear genotype-phenotype correlation due to clinical variability, unspecific symptoms, and high rates of genetic variants restricted to single families.⁴

Pathogenicity of the p.Asp313Tyr variant (syn.: D313Y) has been debated for decades. Initially linked to FD but later associated with enzymatic pseudodeficiency,^5,6 it remains controversial despite reclassification as benign/likely benign by most laboratory results according to American College of Medical Genetics and Genomics/Association of Molecular Pathology (ACMG/AMP) variant classification criteria⁷ and ClinGen-SVI Working Group recommendation⁸ (classification criteria applied by us mentioned in Supplementary Material). Some studies suggest a potential functional impact in cerebrovascular disease or small fiber neuropathy.⁹ To clarify its role, we systematically analyzed a large patient cohort from Germany from our Biodatabank.

Methods

A German cohort (2015–2020) suspected of FD from Centogene's Biodatabank was analyzed for sex, age at first test, genotype, clinical information, and AGLA activity and Lyso-Gb3 value (for male patients only). Observational period and region were selected to assure comparable clinical workflows and analytic techniques. Clinical data were categorized into stroke/TIA, acroparesthesia/neuropathic pain/neuropathy, and cardiac and renal involvement. Genotyping, and enzyme activity and Lyso-Gb3 measurement were performed on dried blood spots (CentoCard) as described previously.¹⁰

Statistical Analysis

Odds ratios of allele frequency were tested using chi-square tests with Yates correction. To adjust for interdependence of clinical symptom domains, we applied multivariable logistic regression for predicting the affiliation to one of the genetic groups compared with the reference cohort. For comparison of biochemical parameters (AGLA activity, Lyso-Gb3 level), the Fisher exact test after Bonferroni correction was applied.

Standard Protocol Approvals, Registrations, and Patient Consents

The research was performed in accordance with the Declaration of Helsinki and conducted within a diagnostic setting, using deidentified data and samples in the second step, thus not requiring institutional review board approval in our jurisdiction. Informed consent was obtained from the referring physicians, patients, and/or their parents or legal guardians. The informed consent form is available in English and several other languages at centogene.com/downloads.html.

Data Availability

Anonymized data not published within this article will be made available by request from any qualified investigator.

Results

Genetic Background

Between 2015 and 2020, 155,410 patient samples underwent GLA gene analysis through targeted sequencing or whole-exome/genome methods. In the German subcohort (38,242 samples; 18,823 female patients/19,419 male patients), 539 individuals (303 female patients/236 male patients) carried pathogenic/likely pathogenic (P/LP) variants, 211 had variants of uncertain significance (VUSs), and 131 had benign/likely benign (B/LB) variants. The D313Y variant was observed in 529 individuals (318 female patients/211 male patients), showing an allele frequency of 0.0093 (0.0084 in female patients/0.0109 in male patients)—twice as high as the gnomAD v4.1.0 European non-Finnish population rate (0.0044; 0.0043 in female patients/0.0045 in male patients).¹¹ This yielded an odds ratio of 2.13 (95% CI 1.94–2.33; p < 0.0001). In 36,832 individuals (17,987 female patients/18,845 male patients), no GLA variant was detected (i.e., reference group) (Figure).

Figure. Flowchart Presenting Selection of the Study Cohort and Genotypic Distribution of All Analyzed Samples Grouped by Reference (No GLA Variant), Any Pathogenic/Likely Pathogenic (P/LP) Variant, Any Variant of Uncertain Significance (VUS), Any Benign/Likely Benign (B/LB) Variant, and p.Asp313Tyr Variant (D313Y).

Clinical Phenotype

Clinical data, available for 27,220 individuals (13,251 female patients /13,969 male patients), showed P/LP variants significantly linked to neuropathy and cardiac and renal symptoms (with reduced stroke/TIA odds as also demonstrated for VUS) compared with the reference group. No symptom associations were found for D313Y or B/LB cohorts (Table 1).

Table 1.

Distribution of Clinical Symptoms Among Different Genotypic Groups Compared With the Reference Cohort

Predictors	Odds ratio	95% CI	p Value
	D313Y variant
Stroke/TIA	1.22	0.82–1.82	0.336
Acroparesthesia/neuropathic pain/neuropathy	1.12	0.69–1.78	0.645
Cardiomyopathy/left ventricular hypertrophy	0.89	0.49–1.49	0.674
Renal involvement	0.84	0.52–1.33	0.481
	P/LP GLA variants
Stroke/TIA	0.32	0.16–0.60	0.001
Acroparesthesia/neuropathic pain/neuropathy	2.65	1.61–4.34	<0.001
Cardiomyopathy/left ventricular hypertrophy	1.65	0.98–2.67	0.047
Renal involvement	2.71	1.67–4.39	<0.001
	VUS
Stroke/TIA	0.38	0.19–0.77	0.007
Acroparesthesia/neuropathic pain/neuropathy	0.51	0.20–1.18	0.135
Cardiomyopathy/left ventricular hypertrophy	0.44	0.11–1.24	0.175
Renal involvement	0.47	0.19–1.06	0.081
	B/LB GLA variants
Stroke/TIA	1.05	0.43–2.62	0.920
Acroparesthesia/neuropathic pain/neuropathy	0.97	0.29–2.75	0.953
Cardiomyopathy/left ventricular hypertrophy	0.27	0.01–1.28	0.197
Renal involvement	0.92	0.30–2.49	0.880

Abbreviations: B/LB = benign/likely benign; P/LP = pathogenic/likely pathogenic; VUS = variant of uncertain significance.

Biochemical Phenotype

Biochemical analysis focused on male patients because of X-inactivation variability in female patients. AGLA activity data included 613 reference samples with no GLA variant, 130 cases with P/LP, 84 cases with D313Y, 46 cases with VUSs, and 30 cases with B/LB GLA variants; Lyso-Gb3 data were available for 744 reference samples, 235 with P/LP, 142 with D313Y, 61 with VUSs, and 39 B/LB cases. AGLA activity ≤35% of normal (deemed clinically significant)¹² differed significantly across groups (Fisher exact test, p < 0.001), as did Lyso-Gb3 (p < 0.001). Substantially reduced AGLA activity was found significantly more often in P/LP variants and in VUSs than in the D313Y variant (Fisher exact test after Bonferroni correction, ps = 0.002; B/LB p = 0.27).

For Lyso-Gb3, significant differences in pairwise comparisons were only found between D313Y and P/LP variants (Fisher exact test after Bonferroni correction, ps = 1.0; P/LP p < 0.001) (Table 2). Normal Lyso-Gb3 was seen in 4 FD cases (p.Arg112His, n = 2; p.Met267Thr; p.Met290Leu).

Table 2.

Distribution of Biochemical Properties (α-Galactosidase A Activity, Globotriaosylsphingosine Level) Among Different Genotypic Groups

Group	AGLA activity (normal value: ≥15.3 µmol/l/hour)			Lyso-Gb3 (normal value: ≤1.8 ng/mL)
	Reduced (≤35% of normal)	Reduced (>35% of normal)	Normal	Increased	Normal
Reference	0	262	351	2	742
D313Y	0	62	22	0	142
P/LP GLA variants	129	1	0	231	4
VUS	16	28	2	0	61
B/LB GLA variants	0	16	14	0	39

Abbreviations: AGLA = α-galactosidase A; B/LB = benign/likely benign; Lyso-Gb3 = globotriaosylsphingosine; P/LP = pathogenic/likely pathogenic; VUS = variant of uncertain significance.

Discussion

The GLA-D313Y variant is classified as benign by ACMG/AMP criteria, consistent with international and all US genetic laboratory results after 2017 as documented in ClinVar.¹³ Controversy arises from studies linking D313Y to FD symptoms in high-risk cohorts (e.g., stroke and cardiomyopathy), but these often lack appropriate controls, risking ascertainment bias and misdiagnosis. Our large diagnostic cohort (38,242 samples) found no significant clinical/biochemical FD associations with D313Y. Although we did not receive clinical information from all patients tested, the rates of clinical data were comparable between D313Y (68%) and reference (72%) groups and there is no reason to assume a selective underreporting of clinical symptoms in specific patient populations regarding the chosen inclusion criteria.

AGLA activity reductions occurred in D313Y but never reached clinically relevant thresholds, mirroring benign variants or reference cases. Furthermore, enzymatic “pseudodeficiency” with D313Y in plasma compared with leukocytes has been repeatedly reported.⁶ A substantial reduction has only been demonstrated in the P/LP cohort and some patients with VUSs. Lyso-Gb3, a key FD biomarker, was normal in all D313Y, VUS, and B/LB cohorts (0% elevated vs 98.3% in P/LP male patients). Again, the relatively high proportion of missing values due to a delayed implementation of a comprehensive standardized workflow including AGLA measurement compared with a “genetics-only” approach in the earlier phase of the FD screening program could potentially confound our results.

The higher D313Y allele frequency in our cohort (OR = 2.13 vs population data) demands further clarification, even if the odds ratio does not allow us to apply the ACMG/AMP classification criterion for PS4. Underreporting of family relations between individuals could have caused an “artificial enrichment” of the variant. However, we could at least correct for obvious (i.e., reported) family relations and included only one individual per family in such cases. This type of bias is often neglected in publications discussing a causative relation between clinical presentation and the D313Y variant.¹⁴

Some studies report mild late-onset phenotypes in D313Y with cerebrovascular or neuropathic involvement,⁹ an observation we could not evidence in our cohort, or even a higher prevalence of D313Y among patients with Parkinson.¹⁵ To clarify the role of D313Y as a potential risk factor or modifier of a, yet to be named, neurologic dysfunction, properly designed case-control studies are needed, an approach beyond the objective of our observational study.

Glossary

AGLA

α-galactosidase A

FD

Fabry disease

LB

likely benign variant

LP

likely pathogenic variant

Lyso-Gb3

globotriaosylsphingosine

Author Contributions

T. Boettcher: drafting/revision of the manuscript for content, including medical writing for content; study concept or design; analysis or interpretation of data. C. Beetz: drafting/revision of the manuscript for content, including medical writing for content; study concept or design. D. Schulze: analysis or interpretation of data. D. Saravanakumar: major role in the acquisition of data; analysis or interpretation of data. E. Zonic: major role in the acquisition of data; analysis or interpretation of data. S. Schroeder: major role in the acquisition of data; analysis or interpretation of data. A. Kaune: major role in the acquisition of data; analysis or interpretation of data. S. Oppermann: major role in the acquisition of data; analysis or interpretation of data. O. Paknia: drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data. J.P. Basto: drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data. P. Bauer: drafting/revision of the manuscript for content, including medical writing for content; study concept or design; analysis or interpretation of data.

Study Funding

The authors report no targeted funding.

Disclosure

T. Boettcher is a part-time employee of Centogene GmbH, holds stock options of Centogene N.V., has served as consultant and received speakers honoraria from Takeda, Amicus and Chiesi. Ch. Beetz is a full-time employee of Centogene GmbH. D. Schulze is a consultant of Centogene GmbH. D. Saravanakumar, E. Zonic, S. Schroeder, A. Kaune, S. Oppermann, O. Paknia, and J. Pinto Basto are full-time employees of Centogene GmbH. P. Bauer is a full-time employee of Centogene GmbH and holds stock options of Centogene N.V. Full disclosure form information provided by the authors is available with the full text of this article at Neurology.org/NG.